INNER CODE UNIT · Python

MRJsonProcessor

ncbi/dbsnp · tutorials/hadoop_json_clinical.py:52

class MRJsonProcessor(MRJob):

    def mapper(self, _, line):

        data = json.loads(line)

        snp_id = data["refsnp_id"]

        annotations = data["primary_snapshot_data"]["allele_annotations"]

        for a in annotations:
            clinical = a["clinical"]

            for c in clinical:
                accession_version = c["accession_version"]
                allele_id = str(c["allele_id"])
                measure_set_id = str(c["measure_set_id"])
                variant_identifiers = c["variant_identifiers"]

View source record →